> Is there any details about the test results? As described in there, 6M poinsts input and Oracle BDA was used. Hama finishes KMeans in a few seconds. Mahout takes almost 500 ~ 1000 secs. See also http://lambda.uta.edu/mrql-bsp.pdf.
Do you need more? On Fri, Jan 3, 2014 at 11:03 AM, Yexi Jiang <[email protected]> wrote: > Is there any details about the test results? > > > 2014/1/2 Edward J. Yoon <[email protected]> > >> Input was too small :/ So, I'll update website using confident benchmarks >> soon. >> >> https://twitter.com/tjungblut/status/414717432293363712 >> >> After I saw this tweet, I thought we need to update the website. Many >> people seems think that hama is a graph processing framework. If you >> have some good idea, Please let me know. >> >> P.S., I recently tested Hama examples and 80% of them didn't work. >> Let's fix them all in 0.7 release and update the website clearly. >> >> >> On Thu, Jan 2, 2014 at 11:55 PM, Tommaso Teofili >> <[email protected]> wrote: >> > Hi all, >> > >> > I just noticed that the graph on our homepage [0] looks very similar to >> the >> > one on Spark homepage [1] so I wonder if we could at least make it a bit >> > clearer either by writing the benchmarks results in a table near it (it >> > seems Hama always takes ~0) or something else I cannot think to right >> now. >> > >> > The reference to the benchmarks wiki page is ok but I cannot find the >> entry >> > for the comparison with Mahout, maybe I'm missing something ... >> > Regards and happy new year everyone. >> > Tommaso >> > >> > [0] : http://hama.apache.org/images/mahout_vs_hama.png >> > [1] : http://spark.incubator.apache.org/images/spark-lr.png >> > >> > >> > 2013/12/20 Edward J. Yoon <[email protected]> >> > >> >> Hi all, >> >> >> >> I published new website for our community. If you have other ideas, >> please >> >> feel free to share your comments or file a JIRA ticket. >> >> >> >> >> >> On Fri, Dec 20, 2013 at 11:29 AM, Edward J. Yoon <[email protected] >> >> >wrote: >> >> >> >> > Thanks, I'll. >> >> > >> >> > Sent from my iPhone >> >> > >> >> > > On 2013. 12. 19., at 오후 10:41, Yexi Jiang <[email protected]> >> wrote: >> >> > > >> >> > > Hi, >> >> > > >> >> > > It looks nice. Is it possible to add more description to the figure? >> >> When >> >> > > people first saw this, they may not know what the x axis is (the >> number >> >> > of >> >> > > cores or the number of the number of groom servers?). Moreover, it >> is >> >> > > better to tell the reader some specs of the dataset used. >> >> > > >> >> > > Regards, >> >> > > Yexi >> >> > > >> >> > > >> >> > > 2013/12/19 Edward J. Yoon <[email protected]> >> >> > > >> >> > >> Thank you so much! >> >> > >> >> >> > >> Sent from my iPhone >> >> > >> >> >> > >>>> On 2013. 12. 19., at 오후 4:45, Tommaso Teofili < >> >> > [email protected]> >> >> > >>> wrote: >> >> > >>> >> >> > >>> Hi Edward, >> >> > >>> >> >> > >>> I think it generally looks better than the current one, I would >> just >> >> > >> change >> >> > >>> this: >> >> > >>> >> >> > >>> Many data analysis techniques such as machine learning and graph >> >> > >> algorithms >> >> > >>> require iterative computations but MapReduce model doesn't fit for >> >> > these >> >> > >>> iterative data analysis applications. To run these iterative data >> >> > >> analysis >> >> > >>> applications more efficiently, Hama offers pure Bulk Synchronous >> >> > Parallel >> >> > >>> computing engine. >> >> > >>> >> >> > >>> >> >> > >>> to something like this: >> >> > >>> >> >> > >>> Many data analysis techniques such as machine learning and graph >> >> > >> algorithms >> >> > >>> require iterative computations, this is where Bulk Synchronous >> >> Parallel >> >> > >>> model can be more effective than "plain" MapReduce. Therefore to >> run >> >> > such >> >> > >>> iterative data analysis applications more efficiently, Hama offers >> >> pure >> >> > >>> Bulk Synchronous Parallel computing engine. >> >> > >>> >> >> > >>> >> >> > >>> As I wouldn't say MR is inherently not good for iterative >> >> computations, >> >> > >>> just BSP can be a better / more perfomant alternative. >> >> > >>> My 2 cents, >> >> > >>> Tommaso >> >> > >>> >> >> > >>> >> >> > >>> 2013/12/19 Edward J. Yoon <[email protected]> >> >> > >>> >> >> > >>>> Hi, >> >> > >>>> >> >> > >>>> I've made some changes to our website - >> >> > >>>> http://people.apache.org/~edwardyoon/site/ - Please review and >> >> > feedback >> >> > >>>> here. >> >> > >>>> >> >> > >>>> -- >> >> > >>>> Best Regards, Edward J. Yoon >> >> > >>>> @eddieyoon >> >> > > >> >> > > >> >> > > >> >> > > -- >> >> > > ------ >> >> > > Yexi Jiang, >> >> > > ECS 251, [email protected] >> >> > > School of Computer and Information Science, >> >> > > Florida International University >> >> > > Homepage: http://users.cis.fiu.edu/~yjian004/ >> >> > >> >> >> >> >> >> >> >> -- >> >> Best Regards, Edward J. Yoon >> >> @eddieyoon >> >> >> >> >> >> -- >> Best Regards, Edward J. Yoon >> @eddieyoon >> > > > > -- > ------ > Yexi Jiang, > ECS 251, [email protected] > School of Computer and Information Science, > Florida International University > Homepage: http://users.cis.fiu.edu/~yjian004/ -- Best Regards, Edward J. Yoon @eddieyoon
